Cami - Claim Assistant AI

CAMI - Claim Assistant AI

Inspiration

The inspiration for Cami came directly from a recent personal experience. When my wife and I welcomed our new baby (thank you!), we quickly discovered how incredibly manual and time-consuming insurance claim submissions can be. What should have taken minutes stretched into days and even weeks due to manual work. I realized agentic AI can solve so much of this type of tedious tasks, and give better experience overall allowing customers to have natural human like conversations to accomplish boring workflows.

What it does

Cami helps you manage and submit insurance claims more easily. It takes over the repetitive parts, like filling out forms in more human conversational way. Cami also guides you through the process, making sure you understand what's needed and helping you correctly input details. If you're unsure about your policy, it can explain relevant sections, and it streamlines the task of submitting bills. Basically, it's designed to cut down the manual work and confusion involved in submitting claims.

How we built it

We used Google's ADK and support for multi-agent workflows and lots of prompt engineering.

Challenges we ran into

Designing agents requires a lot of prompt engineering with multiple iterations and understanding human workflows.

Accomplishments that we're proud of

We liked the way the multi agent workflows handles real world scenarios.

What we learned

Agent tools and descriptions are as important as llm prompt engineering, keeping the tools simple yet descriptive can help the llm navigate the nuances better.